Real-time early warning algorithm and system for slope deformation based on deep reinforcement learning

Abstract

The present invention relates to the field of deep learning technology, and in particular provides a real-time early warning algorithm and system for slope deformation based on deep reinforcement learning. The method includes generating a standardized state representation vector and establishing a state quality evaluation index; receiving the state representation vector as input, constructing a dual-mode strategy framework, generating a preliminary decision and its confidence evaluation, and forming a quantitative indicator of decision reliability; forming a final early warning instruction and strategy update suggestion, and generating a strategy performance evaluation report; continuously tracking the changing trend of the state feature distribution, constructing a strategy degradation early warning system, and triggering a model reconstruction mechanism; the updated strategy network parameters are fed back to the dual-mode strategy framework, and the optimized feature extraction suggestion is fed back to establish a state quality evaluation index. The system includes an indicator establishment module, a report generation module, and a reconstruction mechanism module. The real-time early warning algorithm for slope deformation of the present invention realizes an end-to-end closed-loop control system of monitoring-decision-making-optimization through a deep reinforcement learning framework.

Description

Technical Field

[0001] The present invention relates to the field of deep learning technology, and in particular to a real-time early warning algorithm and system for slope deformation based on deep reinforcement learning. Background Art

[0002] With the rapid development of infrastructure construction in my country, stability monitoring of slopes in projects such as highways, railways, and water conservancy projects has become a major issue in the public safety field. Traditional slope monitoring relies primarily on periodic manual inspections such as GNSS measurements and fixed-point observations with total stations. This is plagued by long data collection intervals and significant response delays. Existing systems struggle to capture critical slope deformation characteristics in a timely manner, particularly during sudden geological disasters such as rainstorms and earthquakes. This results in a high proportion of slope accidents nationwide caused by delayed warnings. Although InSAR remote sensing technology has expanded its monitoring range in recent years, its centimeter-level accuracy still cannot meet the early warning requirements for millimeter-level deformations on high-risk slopes.

[0003] Prior art one, Chinese patent, application number: CN202411648740.6 discloses a real-time monitoring and early warning device for deformation monitoring slopes based on millimeter wave radar, which belongs to the field of real-time monitoring and early warning technology for slopes. It is provided with a monitoring positioning seat that is easy to position and assemble, and a support seat is installed on the upper surface of the positioning seat; it includes: a control board, which is installed on the inner surface of the support seat, and the upper surface of the inner surface of the support seat is rotatably connected to a worm, and the outer surface of the worm is meshedly connected to a worm wheel, and a rotating part is installed on the upper surface of the worm wheel. A rotating monitoring mechanism is provided to effectively control the position of the positioning seat, and cooperate with the support seat and worm assembled by the positioning seat, as well as the worm wheel and the rotating part to control the angle of the monitoring rod assembled by the connecting seat, thereby controlling the angle of the monitoring head, controlling the monitoring range, and coordinating the angle adjustment of the monitoring head and the monitoring rod. Although the situation of the opposite surface slope is scanned according to the assembly height of the positioning seat, thereby controlling the ease of use of the monitoring head; however, there are limitations on the monitoring dimension caused by the single sensor data source (millimeter wave radar); the inherent response delay defect of mechanical scanning monitoring (worm gear angle adjustment); and the bottleneck of the traditional equipment control system (positioning seat-support seat mechanical structure) lacking autonomous decision-making capabilities.

[0004] Prior art 2, Chinese patent, application number: CN202011422338.8 discloses an open-pit mine slope deformation measurement method integrating InSAR and GNSS, which includes two technical means: synthetic aperture radar interferometry InSAR and global navigation satellite system GNSS, and includes the following steps: using the InSAR monitoring method to obtain preliminary deformation monitoring results of the slope; deploying GNSS online monitoring points at the profile position passing through the deformation center; using the least squares iteration method to obtain the corrected instantaneous deformation field; using the adaptive filtering method, Kalman filter equation group and Kriging interpolation method to calculate and obtain deformation monitoring results with full coverage of the time and space domain. Although the fusion of InSAR and GNSS's advantages in spatial and temporal resolution can complement each other, it can achieve both precise point-based real-time monitoring of key areas on open-pit mine slopes and comprehensive surface monitoring of the entire mining area, providing support for real-time measurement and early warning of open-pit mine slope deformation; however, the monitoring gaps caused by the mismatch in spatiotemporal resolution between InSAR and GNSS fusion measurements, the strong dependence of traditional algorithms such as least squares iteration on historical data, and the insufficient applicability of spatial interpolation methods such as Kriging interpolation in real-time early warning scenarios.

[0005] Currently, existing technologies 1 and 2 suffer from problems such as insufficient multi-source data fusion, poor coordination between mechanical structures and electronic systems, difficulty matching spatiotemporal resolution, and insufficient real-time performance of traditional algorithms. Therefore, the present invention provides a real-time slope deformation warning algorithm and system based on deep reinforcement learning. Summary of the Invention

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] In one aspect of the present invention, a real-time early warning algorithm for slope deformation based on deep reinforcement learning is provided, comprising the following steps:

[0008] Acquire multi-source monitoring data that has been screened for state characteristics, generate standardized state representation vectors, and establish state quality assessment indicators;

[0009] Receive the state representation vector as input, build a dual-mode strategy architecture, generate preliminary decisions and their confidence assessments, and form quantitative indicators of decision reliability; form final warning instructions and strategy update recommendations, and generate a strategy performance evaluation report;

[0010] Based on the strategy performance evaluation report, identify the weak links of the strategy; continuously track the changing trend of the state feature distribution, build a strategy degradation warning system, and trigger the model reconstruction mechanism; the updated strategy network parameters are fed back to the dual-mode strategy architecture, and the optimized feature extraction suggestions are fed back to establish the state quality evaluation indicators.

[0011] In an optional implementation, multi-source monitoring data of displacement, strain and groundwater level are integrated through a dynamic weighted fusion mechanism, and state characteristics of the multi-source monitoring data are screened.

[0012] In an optional implementation, the process of establishing a state quality assessment indicator includes the following steps:

[0013] Receive data streams of real-time multi-source monitoring data from displacement sensors, strain gauges, and groundwater level gauges; calculate preliminary weighting coefficients based on the historical drift rates of displacement sensors, strain gauges, and groundwater level gauges; use moving averages to detect sudden changes in single-source data; and mark abnormal multi-source monitoring data;

[0014] The displacement change rate and strain increment in the data stream are synchronized in time and spatially matched with the groundwater level change data; a dynamic correlation coefficient matrix is constructed to determine whether the coordinated change trend of the displacement sensor, strain gauge and groundwater level gauge data matches the physical mechanism;

[0015] The data stream that matches the physical mechanism is compressed through principal component analysis to generate a low-dimensional state representation vector; the statistical dispersion of the low-dimensional state feature vector in a recent time window is calculated to monitor the distribution offset of multi-source monitoring data in the data stream.

[0016] In an optional implementation, the process of generating a policy performance evaluation report includes the following steps:

[0017] Receive the real-time warning instruction sequence output by the dual-mode strategy architecture and perform spatiotemporal alignment verification with the slope deformation events that actually occurred during the same period; obtain the true positive and false positive event matching rates within the instruction time window to quantify the time series detection sensitivity; obtain the position detection accuracy through spatial overlap analysis to form an initial detection effectiveness matrix;

[0018] Based on the decision confidence distribution of 30 consecutive monitoring cycles, a strategy volatility index is constructed. At the same time, the trend of the strategy volatility index value under different geological conditions is analyzed to identify environmental sensitivity parameters and generate a strategy robustness surface.

[0019] Extract the key feature weight vectors from the state quality assessment indicators and perform reverse mapping with the decision error event set; establish a feature-error correlation matrix; obtain the main failure modes through singular value decomposition and mark the feature combinations that need to be optimized; construct a three-dimensional evaluation coordinate system: the horizontal axis is the immediate detection efficiency, the vertical axis is the long-term stability, and the depth axis is the evolutionary capability. Use Monte Carlo sampling to predict the joint distribution of the three indicators and divide the strategy health level.

[0020] In an optional implementation, the process of receiving the real-time warning instruction sequence output by the dual-mode strategy architecture includes the following steps:

[0021] Input the generated standardized state representation vector and load the state quality assessment index as a verification benchmark. Remove abnormal feature values based on the state quality assessment index and input the verification benchmark into the main decision module and auxiliary verification module of the dual-mode strategy architecture.

[0022] Generates primary warning instructions based on the current state features of the standardized state representation vector and outputs a decision confidence score. Verifies the consistency of the output of the main decision module through feature space projection. If it deviates from the historical robust decision region, a correction coefficient is generated. The confidence score of the main decision module verified by feature space projection and the correction coefficient of the auxiliary module together constitute the decision reliability index.

[0023] The primary warning instructions are dynamically adjusted based on the decision reliability index. If it is greater than the preset threshold, the main decision result is directly output as the final warning instruction. If it is not greater than the preset threshold, the auxiliary correction coefficient is enabled to reconstruct the warning parameters and generate conservative warning instructions. The adjusted warning instructions need to be sent back to the status quality assessment indicator to verify the coverage of historical failure cases.

[0024] In an optional implementation, the construction process of the main decision module and the auxiliary verification module of the dual-mode policy architecture includes the following steps:

[0025] The main decision module input receives the filtered standardized state representation vector as the core feature input and loads the pre-trained policy network weights. The auxiliary verification module input simultaneously receives the same standardized state representation vector and additionally accesses the robust decision boundary parameters constructed from the historical case dataset.

[0026] The main decision module generates a primary warning instruction based on the state feature vector (, and simultaneously calculates the matching degree of the current decision with the historical optimal solution in the feature space and outputs a confidence score; the confidence score reflects the degree to which the current decision deviates from the training data distribution; the auxiliary verification module maps the main decision output to the historical robust decision region through feature space projection, determines whether it exceeds the credible interval, and if so, generates a correction coefficient based on the preset decision robustness rules;

[0027] The confidence score output by the main decision module and the correction coefficient of the auxiliary module are weighted and fused to form the decision reliability index.

[0028] In an optional embodiment, the process of marking the feature combination to be optimized includes the following steps:

[0029] The feature weights in the original state quality assessment indicators are rescaled and standardized. Historical strategy failure cases are categorized according to their performance characteristics to establish a structured error event library. Each category of event is associated with a complete snapshot of the feature parameters at the time of the incident.

[0030] A feature-error correlation matrix is constructed. By calculating the matching degree between feature vectors and error cases, the influence of each feature indicator on each type of error is quantified. The values in the matrix intuitively reflect the strength of the association between a specific feature and a certain type of error. The complete feature-error correlation matrix is spatially compressed and transformed to retain the dominant direction that explains most of the data variation, achieving the transformation from high-dimensional features to core features.

[0031] In the feature space after spatial compression transformation, the main failure modes are screened according to the explanatory power of each dimension; the key dimensions obtained by the screening are reversely mapped back to the original feature space to identify the most influential feature combination pattern that causes strategy failure.

[0032] In an optional implementation, a feature improvement priority list is established based on the failure mode analysis results, and the feature combinations that need to be optimized and their improvement order are clarified through quantitative evaluation of the impact of each mode.

[0033] In an optional implementation, the process of identifying policy weaknesses includes the following steps:

[0034] Receive the early warning strategy performance evaluation report. If the key indicators of the early warning strategy performance evaluation report, such as false alarm rate, confidence matching degree and feature drift index, exceed the preset security threshold, the weak link analysis mechanism will be triggered.

[0035] Extract all false positive state representation vectors, classify them into potential failure modes, compare the data distribution characteristics of each failure cluster, and screen out key feature deviation patterns that are significantly correlated with the current strategy decision error. Calculate the cumulative deviation trend strength of historical monitoring data based on the feature drift index to predict the probability of future strategy failure.

[0036] Based on the failure mode and the predicted probability of future strategy failure, locate the feature set that causes a sudden drop in decision reliability. If the false alarm rate caused by a key feature changes significantly, the key feature is listed as the highest priority for optimization.

[0037] Another aspect of the present invention provides a real-time early warning system for slope deformation based on deep reinforcement learning, comprising:

[0038] The indicator establishment module is used to integrate multi-source monitoring data of displacement, strain and groundwater level through a dynamic weighted fusion mechanism, screen the state characteristics of the multi-source monitoring data, generate standardized state representation vectors, and establish state quality assessment indicators;

[0039] The report generation module is used to receive the state representation vector as input, build a dual-mode strategy architecture, generate preliminary decisions and their confidence assessments, and form quantitative indicators of decision reliability; form final warning instructions and strategy update recommendations, and generate a strategy performance evaluation report;

[0040] The reconstruction mechanism module is used to identify strategy weaknesses based on strategy performance evaluation reports; continuously track the changing trends of state feature distribution, build a strategy degradation warning system, and trigger the model reconstruction mechanism; the updated strategy network parameters are fed back to the dual-mode strategy architecture, and the optimized feature extraction suggestions are fed back to establish state quality evaluation indicators.

[0041] The proposed real-time slope deformation early warning algorithm utilizes a deep reinforcement learning framework to implement an end-to-end closed-loop control system encompassing monitoring, decision-making, and optimization. Its key technical benefits include: The coordinated optimization of dynamic perception and adaptive decision-making. The environmental perception module, constructed through a dynamic weighted fusion mechanism and multi-level state feature screening, forms a closed-loop feedback loop with the dual-mode policy architecture. The quality assessment indicators of the state representation vector guide the policy architecture's mode switching in real time, while policy performance evaluation inversely optimizes feature extraction thresholds, enabling the system to adapt to dynamic environments. The algorithm achieves a balanced control between real-time response and long-term evolution. In the short term, the dual-mode policy architecture achieves millisecond-level early warning decisions based on confidence assessment. In the long term, the policy degradation early warning system continuously tracks changes in the distribution of state features to establish a predictive maintenance mechanism. These two systems collaborate through a parameter update channel. The algorithm deeply integrates data-driven and mechanism-constrained approaches. The algorithm constructs a unified feature space by standardizing the state representation vector, allowing monitoring data at different spatiotemporal scales to be treated as identically distributed samples. Policy update recommendations are generated based on feature-decision correlation analysis, maintaining data-driven characteristics while mitigating overfitting risks through quality control indicators. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0043] Figure 1 This is a flow chart of the real-time early warning algorithm for slope deformation based on deep reinforcement learning provided in Example 1 of the present invention;

[0044] Figure 2 A process diagram for establishing a state quality assessment indicator provided in Example 2 of the present invention;

[0045] Figure 3 A process diagram for generating a strategy performance evaluation report provided in Example 3 of the present invention;

[0046] Figure 4 This is a process diagram for identifying weak links in a strategy provided in Example 7 of the present invention;

[0047] Figure 5 This is a block diagram of a real-time early warning system for slope deformation based on deep reinforcement learning provided in Example 8 of the present invention; Figure 5In: 1. Indicator establishment module; 2. Report generation module; 3. Reconstruction mechanism module;

[0048] Figure 6 A block diagram of the electronic device provided by the present invention; Figure 6 In: 4. CPU / microprocessor / main control chip, etc.; 5. Storage medium; 6. Data bus; 7. Input / output bus / external bus / device bus, etc.; 8. Display; 9. Input / output device;

[0049] Figure 7 A block diagram of a computer-readable storage medium provided by the present invention; Figure 7 In: 10. Instructions; 11. Computer-readable storage medium. DETAILED DESCRIPTION

[0050] The technical solutions in the embodiments of the present invention will be described below with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0051] In the following, the terms "first," "second," etc., are used for descriptive convenience only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature identified with "first," "second," etc., may explicitly or implicitly include one or more of such features. In the description of the present invention, unless otherwise specified, "plurality" means two or more.

[0052] In the present invention, unless otherwise clearly specified and limited, the term "connection" should be understood in a broad sense. For example, "connection" can be a fixed mechanical connection, a detachable mechanical connection, or an integrated one; or, "connection" can be a direct connection or an indirect connection through an intermediate medium. In addition, unless otherwise clearly specified and limited, the term "coupling" should be understood in a broad sense. For example, "coupling" can be a direct electrical connection, such as physical contact and electrical conduction between two components, or it can be understood as the electrical connection between different components in a circuit structure through a physical line that can transmit electrical signals, such as printed circuit board (PCB) copper foil or wire, so as to transmit electrical signals; or, "coupling" can be an indirect electrical connection between two components through an intermediate medium; or, "coupling" can be an electrical connection between two components in an airless / non-contact manner, such as electrical connection between two components using capacitive coupling to transmit electrical signals.

[0053] In an embodiment of the present invention, directional terms such as "up", "down", "left" and "right" may be defined including but not limited to the orientation relative to the schematic placement of the components in the drawings. It should be understood that these directional terms may be relative concepts, which are used for relative description and clarification, and may change accordingly according to changes in the orientation of the components in the drawings.

[0054] Example 1:

[0055] like Figure 1 As shown, an embodiment of the present invention provides a real-time early warning algorithm for slope deformation based on deep reinforcement learning, which includes the following steps:

[0056] S100: Integrates multi-source monitoring data such as displacement, strain, and groundwater level through a dynamic weighted fusion mechanism, screens the state characteristics of the multi-source monitoring data, generates standardized state representation vectors, and establishes state quality assessment indicators;

[0057] S200: Receive the state representation vector as input, build a dual-mode strategy architecture, generate a preliminary decision and its confidence assessment, and form a quantitative indicator of decision reliability; form a final warning instruction and strategy update suggestion, and generate a strategy performance evaluation report;

[0058] S300: Based on the strategy performance evaluation report, identify strategy weaknesses; continuously track the changing trend of state feature distribution, build a strategy degradation warning system, and trigger the model reconstruction mechanism; the updated strategy network parameters are fed back to the dual-mode strategy architecture, and the optimized feature extraction suggestions are fed back to establish state quality evaluation indicators.

[0059] In the aforementioned embodiment, the real-time slope deformation warning algorithm of this embodiment implements an end-to-end closed-loop control system of monitoring, decision-making, and optimization through a deep reinforcement learning framework. Its technical benefits are primarily reflected in the coordinated optimization of dynamic perception and adaptive decision-making. The environmental perception module, constructed through a dynamic weighted fusion mechanism and multi-level state feature screening, forms a closed-loop feedback loop with the dual-mode policy architecture. The quality assessment indicators of the state representation vector guide the policy architecture's mode switching in real time, while the policy performance evaluation inversely optimizes the feature extraction threshold, enabling the system to adapt to dynamic environments. A balanced control strategy of real-time response and long-term evolution is achieved. In the short term, the dual-mode policy architecture achieves millisecond-level warning decisions based on confidence assessment. In the long term, the policy degradation warning system continuously tracks changes in the distribution of state features to establish a predictive maintenance mechanism. These two systems collaborate through a parameter update channel. The algorithm deeply integrates data-driven and mechanism-constrained approaches. The algorithm constructs a unified feature space by standardizing the state representation vector, allowing monitoring data at different temporal and spatial scales to be treated as identically distributed samples. Policy update recommendations are generated based on feature-decision correlation analysis, maintaining data-driven characteristics while mitigating overfitting risks through quality control indicators.

[0060] Example 2:

[0061] like Figure 2 As shown, based on Example 1, the process of establishing the state quality assessment index in S100 provided in this embodiment of the present invention includes the following steps:

[0062] S101: Receive data streams of real-time multi-source monitoring data from displacement sensors, strain gauges, and groundwater level gauges; calculate preliminary weighting coefficients based on historical drift rates of the displacement sensors, strain gauges, and groundwater level gauges; use a moving average to detect sudden changes in single-source data; and mark abnormal multi-source monitoring data.

[0063] S102: Time synchronization and spatial matching of the displacement change rate and strain increment in the data stream with the groundwater level change data; constructing a dynamic correlation coefficient matrix to determine whether the coordinated change trend of the displacement sensor, strain gauge, and groundwater level gauge data matches the physical mechanism;

[0064] S103: Compress the data stream that matches the physical mechanism through principal component analysis to generate a low-dimensional state representation vector; calculate the statistical dispersion of the low-dimensional state feature vector in a recent time window to monitor the distribution deviation of multi-source monitoring data in the data stream.

[0065] Among them, the initial weighting coefficient of S101 and the abnormal mark, dynamic drift rate correction weighting coefficient expression are:

[0066]

[0067] Where, represents the weighting coefficient of sensor i at time t; and Represents the sensor i and j at the historical moment drift rate (long-term measurement error mean); represents the drift difference sensitivity parameter; represents the length of the sliding time window (the unit represents the number of sampling points); N represents the total number of sensors; represents the mutation penalty factor; represents the instantaneous rate of change of sensor i at time t (first-order difference); Display window Moving average within ; Natural exponential function; Sign function, outputs the positive or negative value of the input value; Represents sensor k at the historical moment Drift rate; t current moment;

[0068] Multi-source data mutation detection threshold expression:

[0069]

[0070] Where, represents the abnormal flag of sensor i at time t (1 means abnormal); Indicates the standard deviation multiple threshold; Indicates a numerical stability constant; if indicates a conditional statement; represents the data differential signal; k represents the sliding pointer in the time window;

[0071] S102 represents the verification of spatiotemporal matching and dynamic correlation coefficient, and the data resampling function expression after spatiotemporal alignment is:

[0072]

[0073] Where, Represents sensor i at the time and space point Interpolation data of , Represents the time and space interpolation weights; , represents the bandwidth parameter of the spatiotemporal Gaussian kernel; represents the target space coordinates, Represents the spatial coordinates of the sensor; Represents sensor i at discrete time stamp The original monitoring data value of Indicates the number of time dimension reference points in spatiotemporal interpolation; represents the time reference point index; m represents the spatial reference point index; M represents the total number of spatial reference points;

[0074] Dynamic correlation coefficient matrix and physical mechanism deviation expression: Indicates the spatial position of sensor i Monitoring data value;

[0075]

[0076]

[0077] Where, Indicates that sensors i and j are in the window Dynamic correlation coefficient within ; represents the expected correlation coefficient based on the physical mechanism; It represents the overall mechanism deviation index; represents the time gradient attenuation factor; represents the interpolated data of sensor i after spatial and temporal alignment at historical time τ; represents the interpolated data of sensor i in the time window [t−T,t] The mean of

[0078] S103 principal component compression and distribution shift monitoring, time-varying covariance matrix and principal component projection expression:

[0079]

[0080]

[0081] Where, represents the regularized covariance matrix; express forward The projection matrix composed of eigenvectors; Represents the low-dimensional state representation vector; T represents the width of the time window;

[0082] Distribution shift monitoring indicator expression:

[0083]

[0084] Where, represents the regularized covariance matrix The lth eigenvalue of ; Represents the history window Inside The mean of Indicates the history window The standard deviation of express The lth eigenvector of ; H represents the length of the historical reference window;

[0085] Explanation of logical consistency between formulas: S101 uses dynamic drift rate to correct weighting coefficients (and mutation detection to screen reliable data streams). S102 uses time series interpolation to align data and verifies physical consistency through dynamic correlation coefficients. S103 uses time-varying covariance matrix dimensionality reduction and quantifies data distribution changes through distribution offset indicators. All formulas form a closed-loop feedback loop through parameters to ensure system adaptive optimization.

[0086] In the aforementioned embodiments, this embodiment implements a comprehensive system for optimizing and characterizing slope deformation monitoring data. The system uses weighting coefficients and mutation detection in historical drift rate calculations to ensure the reliability of input data streams and reduce the impact of noise or abnormal data on subsequent analysis. Dynamic weighting and correlation analysis are combined to select multi-source data with consistent physical mechanisms, avoiding computational bias caused by single sensor failure or environmental interference. Time synchronization and spatial matching are used to eliminate monitoring data bias caused by sampling intervals or placement, ensuring the effectiveness of collaborative analysis of displacement, strain, and groundwater level data. A dynamic correlation coefficient matrix quantifies the correlation of sensor data, eliminating data that deviates from the overall slope deformation trend due to local interference (such as groundwater level fluctuations caused by short-term rainfall). Principal component analysis transforms high-dimensional, unstructured multi-source monitoring data into compact, low-dimensional state representation vectors, reducing computational complexity while preserving key physical characteristics. Distribution shift monitoring based on statistical dispersion calculations provides real-time perception of data stream trends, avoiding model degradation caused by static modeling. Dynamic weighting adjustments or decision strategy optimization can be directly triggered.

[0087] In summary, this embodiment constructs a comprehensive state quality assessment system for real-time monitoring with adaptive dynamic weight adjustment, physics-driven feature selection, and distribution-sensitive performance. The system ensures the reliability of multi-source sensor data, enhances the availability of monitoring information, and responds to environmental changes or equipment degradation.

[0088] Example 3:

[0089] like Figure 3 As shown, based on Example 1, the process of generating a policy performance evaluation report in S200 provided in this embodiment of the present invention includes the following steps:

[0090] S201: Receive the real-time warning instruction sequence output by the dual-mode strategy architecture, perform spatiotemporal alignment verification on it and the slope deformation events actually occurring during the same period; obtain the true positive and false positive event matching rates within the instruction time window, and quantify the time series detection sensitivity; obtain the position detection accuracy through spatial overlap analysis, and form an initial detection effectiveness matrix;

[0091] S202: Based on the decision confidence distribution of 30 consecutive monitoring cycles, a strategy volatility index is constructed; at the same time, the trend of the strategy volatility index value under different geological conditions is analyzed, the environmental sensitivity parameters are identified, and the strategy robustness surface is generated;

[0092] S203: Extract the key feature weight vectors from the state quality assessment indicators and perform reverse mapping with the decision error event set; establish a feature-error correlation matrix; obtain the main failure modes through singular value decomposition and mark the feature combinations that need to be optimized; construct a three-dimensional evaluation coordinate system: the horizontal axis is the immediate detection efficiency, the vertical axis is the long-term stability, and the depth axis is the evolutionary capability. Use Monte Carlo sampling to predict the joint distribution of the three indicators and divide the strategy health level.

[0093] In the above-mentioned embodiments, first, in establishing a multi-dimensional dynamic verification mechanism, through the closed-loop linkage of the spatiotemporal verification module and state feature backtracking, while simultaneously verifying the warning accuracy in real time (spatiotemporal alignment verification), the mapping relationship between feature vectors and decision errors (feature-error correlation analysis) is dynamically tracked, achieving a deep verification capability from simple result comparison to cause tracing. The verification mechanism can simultaneously complete the error source location in the feature space while maintaining the accuracy of the time window. Secondly, in terms of strategy adaptive optimization, the confidence fluctuation analysis and feature failure mode identification within the continuous monitoring cycle form dual feedback, allowing the system to perceive macroscopic stability changes (robustness surface) and accurately locate microscopic feature defects (main failure mode). The synergistic effect produces a dynamic optimization effect: environmental sensitivity parameters drive coarse-grained strategy adjustments, while feature combination tags guide fine-grained parameter updates. The resulting three-dimensional evaluation system (instant detection effectiveness, long-term stability, and evolutionary capability) achieves a three-dimensional approach to evaluation metrics. Monte Carlo sampling transforms discrete detection data (the initial matrix in S201), continuous stability parameters (the robustness surface in S202), and incremental learning features (the optimized feature set in S203) into a unified probability distribution model. This technical framework extends policy health assessment from traditional single-point static judgments to spatiotemporal evolution predictions, reducing policy lifespan prediction errors compared to traditional methods.

[0094] Example 4:

[0095] Based on Example 3, the process of receiving the real-time warning instruction sequence output by the dual-mode policy architecture in S201 provided in this embodiment of the present invention includes the following steps:

[0096] S2011: Input the generated standardized state representation vector and load the state quality assessment index as a verification benchmark; remove abnormal feature values based on the state quality assessment index, and input the verification benchmark into the main decision module and auxiliary verification module of the dual-mode strategy architecture;

[0097] S2012: Generate a primary warning instruction based on the current state characteristics of the standardized state representation vector and output a decision confidence score. Verify the consistency of the output of the main decision module through feature space projection. If it deviates from the historical robust decision region, generate a correction coefficient. The confidence score of the main decision module verified by feature space projection and the correction coefficient of the auxiliary module together constitute the decision reliability index.

[0098] S2013: Dynamically adjust the primary warning instructions based on the decision reliability index. If it is greater than the preset threshold, the main decision result is directly output as the final warning instruction. If it is not greater than the preset threshold, the auxiliary correction coefficient is enabled to reconstruct the warning parameters and generate a conservative warning instruction. The adjusted warning instruction needs to be sent back to the status quality assessment indicator to verify the coverage of historical failure cases.

[0099] In the above embodiment, this embodiment drives dynamic decision optimization based on a dual verification mechanism of standardized state representation vectors and evaluation indicators. The primary warning instructions are generated by the main decision module while the deviation detection of the auxiliary verification module is executed in parallel. The instructions are then adjusted based on the decision reliability indicators output by the dual modules to form a final warning instruction that is both real-time and robust. The joint input of the state representation vector and the quality evaluation indicator ensures data validity; the correlation coupling of the main module confidence score and the auxiliary module correction coefficient solves the risk of a single decision; the reliability threshold criterion realizes the dynamic switching of the warning response mode; the adjusted warning instructions form a closed-loop feedback through back-transmission verification, enabling the system to balance the warning sensitivity and false alarm rate under the stability constraint of the feature space; reducing the probability of instruction distortion caused by sudden anomalies and improving the coverage of historical failure modes.

[0100] Example 5:

[0101] Based on Example 4, the construction process of the main decision module and the auxiliary verification module of the dual-mode policy architecture in S2011 provided by the embodiment of the present invention includes the following steps:

[0102] S20111: The main decision module input receives the filtered standardized state representation vector as the core feature input and loads the pre-trained policy network weights; the auxiliary verification module input simultaneously receives the same standardized state representation vector and additionally accesses the robust decision boundary parameters constructed from the historical case dataset;

[0103] S20112: The main decision module generates a primary warning instruction based on the state feature vector (, and simultaneously calculates the matching degree of the current decision with the historical optimal solution in the feature space and outputs a confidence score; the confidence score reflects the degree to which the current decision deviates from the training data distribution; the auxiliary verification module maps the main decision output to the historical robust decision region through feature space projection, determines whether it exceeds the credible interval, and if so, generates a correction coefficient based on the preset decision robustness rules;

[0104] S20113: The confidence score output by the main decision module is weighted and fused with the correction coefficient of the auxiliary module to form a decision reliability index.

[0105] In the above-mentioned embodiment, the core technical function of the integrated main decision module and auxiliary verification module of the dual-mode strategy architecture is to achieve dynamic optimization of decision generation through a dual verification mechanism. This architecture, based on standardized state feature vector input, allows the main decision module to output preliminary warning judgments while simultaneously implementing real-time decision verification through the auxiliary verification module, ultimately generating a composite decision output with enhanced reliability. The state representation vector serves as a unified input source to ensure consistency in the operational basis of the two modules. The confidence score generated by the main decision module reflects the degree of deviation between the instantaneous decision and the historical optimal solution, while the spatial projection correction coefficient generated by the auxiliary verification module provides decision boundary constraints. The weighted fusion of the two creates a reliability index that quantifies the confidence of the decision, enabling the system to autonomously switch between the original decision and a robust adjustment scheme based on actual operating conditions. This constructs a self-verifying two-tier decision system. Its technical effectiveness is demonstrated by significantly improving the robustness of decisions under abnormal operating conditions through auxiliary verification while maintaining the real-time responsiveness of the main module. Furthermore, the continuous feedback of the reliability index enables adaptive optimization of system parameters. The entire process forms a closed-loop system from feature input, dual calculation, to decision optimization.

[0106] Example 6:

[0107] Based on Example 3, the process of marking the feature combination to be optimized in S203 provided in the embodiment of the present invention includes the following steps:

[0108] S2031: Standardize the scale of feature weights in the original state quality assessment indicators; categorize historical strategy failure cases according to their performance characteristics and establish a structured error event library. Each category of event is associated with a complete snapshot of feature parameters at the time of the incident.

[0109] S2032: Construct a feature-error correlation matrix. By calculating the matching degree between feature vectors and error cases, the influence of each feature indicator on each error category is quantified. The values in the matrix intuitively reflect the strength of the association between a specific feature and a certain type of error. Perform spatial compression transformation on the complete feature-error correlation matrix to retain the dominant direction that explains most of the data variation, achieving the transformation from high-dimensional features to core features.

[0110] S2033: In the feature space after spatial compression transformation, screen the main failure modes based on the explanatory power of each dimension; reversely map the screened key dimensions back to the original feature space to identify the most influential feature combination patterns that lead to strategy failure; based on the failure mode analysis results, establish a feature improvement priority list, and through quantitative evaluation of the impact of each mode, clarify the feature combinations that need to be optimized and their improvement order.

[0111] In the above embodiment, feature weight normalization and failure case classification establish a standardized feature evaluation system, eliminate the scale differences of original data, build a structured failure case library, and form a failure sample set that can be quantified and analyzed; association modeling and dimensionality compression quantify the degree of correlation between features and failure types, extract the core influencing factors in the high-dimensional feature space, and reduce the computational complexity of the analysis; failure mode identification and optimization guidance identify the feature combinations corresponding to the main failure mechanisms, establish a priority ranking for feature optimization, and provide a basis for improvement direction for strategy iteration;

[0112] In summary, this embodiment systematically transforms raw monitoring data into executable optimization solutions through a progressive process of data standardization, association modeling, dimensionality reduction analysis, and pattern recognition. These steps collaboratively enable structured analysis of feature space and localization and diagnosis of strategic weaknesses, ultimately outputting a statistically robust set of improvement targets.

[0113] Example 7:

[0114] like Figure 4 As shown, based on Example 1, the process of identifying policy weaknesses in S300 provided in this embodiment of the present invention includes the following steps:

[0115] S301: Receive an early warning strategy performance evaluation report. If the key indicators of the early warning strategy performance evaluation report, such as false alarm rate, confidence matching degree, and feature drift index, exceed the preset security threshold, trigger a weak link analysis mechanism.

[0116] S302: Extract all false positive state representation vectors, classify them into potential failure modes, compare the data distribution characteristics of each failure cluster, and screen out key feature deviation patterns that are significantly correlated with the current policy decision error. Calculate the cumulative deviation trend strength of historical monitoring data based on the feature drift index to predict the probability of future policy failure.

[0117] S303: Based on the failure mode and the predicted probability of future strategy failure, locate the feature set that causes a sudden drop in decision reliability. If the false alarm rate caused by a key feature changes significantly, the key feature is listed as the highest priority for optimization.

[0118] In the above-mentioned embodiment, a closed-loop slope deformation early warning system with dynamic adaptability was constructed. Initial feature vectors were generated through multi-source data fusion and standardized representation. A dual-mode strategy architecture was used to make real-time decisions while simultaneously evaluating their credibility. Based on quantitative metrics from the performance evaluation report (false alarm rate, confidence match, and feature drift index), failure modes of the current strategy under specific data distributions were systematically identified (including feature shift cluster analysis and degradation probability modeling). This ultimately triggered the coordinated optimization of strategy parameters and feature selection mechanisms. The overall technical features of this process are: continuous monitoring of strategy performance degradation trends, dynamic identification of decision failures caused by data distribution changes, and parameter iteration to maintain the stability of the early warning system under evolving conditions, achieving a closed-loop operation from data perception to strategy self-correction. The bidirectional feedback mechanism between feature selection optimization and strategy network updates ensures the dynamic compatibility of feature representations with decision logic, effectively reducing the risk of early warning failures due to geological changes or sensor characteristic drift.

[0119] Example 8:

[0120] like Figure 5 As shown, based on Examples 1 to 7, the real-time early warning system for slope deformation based on deep reinforcement learning provided by the embodiment of the present invention includes:

[0121] Indicator establishment module 1 is used to integrate multi-source monitoring data such as displacement, strain, and groundwater level through a dynamic weighted fusion mechanism, screen the state characteristics of the multi-source monitoring data, generate a standardized state representation vector, and establish a state quality assessment indicator;

[0122] Report generation module 2 is used to receive the state representation vector as input, build a dual-mode strategy architecture, generate preliminary decisions and their confidence assessments, and form quantitative indicators of decision reliability; form final warning instructions and strategy update recommendations, and generate a strategy performance evaluation report;

[0123] Reconstruction mechanism module 3 is used to identify strategy weaknesses based on strategy performance evaluation reports; continuously track the changing trends of state feature distribution, build a strategy degradation warning system, and trigger the model reconstruction mechanism; the updated strategy network parameters are fed back to the dual-mode strategy architecture, and the optimized feature extraction suggestions are fed back to establish state quality evaluation indicators.

[0124] In the above-mentioned embodiment, the indicator establishment module uses a dynamic weighted fusion mechanism to transform heterogeneous monitoring data such as displacement, strain, and groundwater level into standardized state representation vectors, eliminating data dimensionality differences and simultaneously screening key state features. This, combined with state quality assessment indicators, enhances data validity and the relevance of feature representation. The report generation module employs a dual-mode strategy architecture, generating confidence assessments simultaneously with warning instructions, providing a quantitative basis for decision reliability. Strategy performance evaluation reports drive strategy updates, achieving a balance between real-time warning and long-term performance optimization in the decision-making system. The reconstruction mechanism module continuously tracks changes in the distribution of state features, detects strategy degradation trends, and triggers model reconstruction. Updated strategy parameters and feature extraction recommendations are fed back to the decision-making layer and data processing layer, respectively, forming a closed-loop optimization chain from feature extraction, decision generation, to model update, ensuring the system's continuous adaptability to dynamic environmental changes. The coupling of data fusion, dynamic decision-making, and a self-correction mechanism enhances the real-time, accuracy, and long-term stability of slope deformation warnings while reducing the risk of false alarms and missed warnings caused by monitoring data noise, environmental changes, or model aging.

[0125] Figure 6 A block diagram is shown of an exemplary electronic device suitable for implementing embodiments of the present invention.

[0126] The electronic device may include a central processing unit / microprocessor / main control chip, etc. 4; a storage medium 5, coupled to the central processing unit / microprocessor / main control chip, etc. 4, and storing computer executable instructions therein for performing the steps of each method of an embodiment of the present invention when executed by the processor.

[0127] The central processing unit / microprocessor / main control chip 4 may include but is not limited to one or more processors or microprocessors.

[0128] The storage medium 5 may include, but is not limited to, for example, random access memory (RAM), read-only memory (ROM), flash memory, EPROM memory, EEPROM memory, registers, computer storage media (such as hard disk, floppy disk, solid-state drive, removable disk, CD-ROM, DVD-ROM, Blu-ray disc, etc.).

[0129] In addition, the electronic device may also include (but not limited to) a data bus 6, an input / output bus / external bus / device bus 7, a display 8, and input / output devices 9 (eg, keyboard, mouse, speaker, etc.).

[0130] The central processing unit / microprocessor / main control chip etc. 4 can communicate with external devices ( 8 , 9 etc.) via an I / O bus 7 via a wired or wireless network (not shown).

[0131] The storage medium 5 may also store at least one computer executable instruction for executing the various functions and / or method steps in the embodiments described in this technology when run by the central processing unit / microprocessor / main control chip 4.

[0132] In one embodiment, the at least one computer executable instruction may also be compiled into or constitute a software product, wherein one or more computer executable instructions are executed by a processor to perform the various functions and / or method steps in the embodiments described in the present technology.

[0133] Figure 7 A schematic diagram of a computer-readable storage medium according to an embodiment of the present invention is shown.

[0134] like Figure 7 As shown, a non-transitory computer-readable storage medium 11 stores instructions, such as computer-readable instructions 10. When the computer-readable instructions 10 are executed by a processor, the various methods described above can be executed. Non-transitory computer-readable storage media include, but are not limited to, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and / or cache memory (cache), etc. Non-transitory non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. For example, the non-transitory computer-readable storage medium 11 can be connected to a computing device such as a computer, and then, when the computing device executes the computer-readable instructions 10 stored on the computer-readable storage medium 11, the various methods described above can be performed.

[0135] In the several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be through some interface, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0136] Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0137] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0138] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for executing all or part of the steps of the various embodiments of the method of the present invention via a computer device (which can be a personal computer, server, or network device, etc.). The aforementioned storage medium includes various media that can store program code, such as USB flash drives, mobile hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0139] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A real-time early warning algorithm for slope deformation based on deep reinforcement learning, characterized by: The following steps are involved: Acquire multi-source monitoring data that has been screened for state characteristics, generate standardized state representation vectors, and establish state quality assessment indicators; It receives the standardized state representation vector as input, builds a dual-mode strategy architecture, generates preliminary decisions and their confidence assessments, and constitutes a quantitative indicator of decision reliability; Form final warning instructions and strategy update suggestions, and generate strategy performance evaluation reports; Identify strategic weaknesses based on the strategy performance evaluation report; Continuously track the changing trends of state feature distribution, build a policy degradation warning system, and trigger the model reconstruction mechanism; the updated policy network parameters are fed back to the dual-mode policy architecture, and the optimized feature extraction suggestions are fed back to establish state quality assessment indicators; The multi-source monitoring data of displacement, strain and groundwater level are integrated through a dynamic weighted fusion mechanism, and the state characteristics of the multi-source monitoring data are screened; The process of establishing status quality assessment indicators includes the following steps: Receive data streams of real-time multi-source monitoring data from displacement sensors, strain gauges, and groundwater level gauges; calculate preliminary weighting coefficients based on the historical drift rates of displacement sensors, strain gauges, and groundwater level gauges; use moving averages to detect sudden changes in single-source data; and mark abnormal multi-source monitoring data; The displacement change rate and strain increment in the data stream are synchronized in time and spatially matched with the groundwater level change data; a dynamic correlation coefficient matrix is constructed to determine whether the coordinated change trend of the displacement sensor, strain gauge and groundwater level gauge data matches the physical mechanism; The data stream that matches the physical mechanism is compressed through principal component analysis to generate a low-dimensional state feature vector; the statistical dispersion of the low-dimensional state feature vector in a recent time window is calculated to monitor the distribution offset of multi-source monitoring data in the data stream.

2. The real-time early warning algorithm for slope deformation based on deep reinforcement learning according to claim 1 is characterized in that: The process of generating a policy performance evaluation report includes the following steps: Receive the real-time warning instruction sequence output by the dual-mode strategy architecture and perform spatiotemporal alignment verification with the actual slope deformation events that occurred during the same period; Obtain the true positive and false positive event matching rates within the instruction time window to quantify the timing detection sensitivity; obtain the position detection accuracy through spatial overlap analysis to form an initial detection efficiency matrix; Based on the decision confidence distribution of 30 consecutive monitoring cycles, a strategy volatility index is constructed. At the same time, the trend of the strategy volatility index under different geological conditions is analyzed to identify environmental sensitivity parameters and generate a strategy robustness surface. Extract the key feature weight vectors from the state quality assessment indicators and perform reverse mapping with the decision error event set; Establish a feature-fault correlation matrix; obtain the main failure modes through singular value decomposition and mark the feature combinations that need to be optimized; construct a three-dimensional evaluation coordinate system: the horizontal axis is immediate detection efficiency, the vertical axis is long-term stability, and the depth axis is evolutionary capability. Use Monte Carlo sampling to predict the joint distribution of the three indicators and divide the strategy health level.

3. The real-time early warning algorithm for slope deformation based on deep reinforcement learning according to claim 2 is characterized in that: The process of receiving the real-time warning instruction sequence output by the dual-mode strategy architecture includes the following steps: Input the generated standardized state representation vector and load the state quality assessment index as a verification benchmark. Remove abnormal feature values based on the state quality assessment index and input the verification benchmark into the main decision module and auxiliary verification module of the dual-mode strategy architecture. Generate primary warning instructions based on the current state characteristics of the standardized state representation vector and output a decision confidence score; The output consistency of the main decision module is verified through feature space projection. If it deviates from the historical robust decision area, a correction coefficient is generated. The confidence score of the main decision module verified by feature space projection and the correction coefficient together constitute the decision reliability index. Dynamically adjust the primary warning instructions based on the decision reliability index. If it is greater than the preset threshold, directly output the main decision result as the final warning instruction; When the value is not greater than the preset threshold, the correction coefficient is activated to reconstruct the warning parameters and generate conservative warning instructions; the conservative warning instructions need to be fed back to the status quality assessment indicator to verify the coverage of historical failure cases.

4. The real-time early warning algorithm for slope deformation based on deep reinforcement learning according to claim 3 is characterized in that: The construction process of the main decision module and auxiliary verification module of the dual-mode strategy architecture includes the following steps: The main decision module input receives the filtered standardized state representation vector as the core feature input and loads the pre-trained policy network weights. The auxiliary verification module input simultaneously receives the same standardized state representation vector and additionally accesses the robust decision boundary parameters constructed from the historical case dataset. The main decision module generates primary warning instructions based on the standardized state representation vector, calculates the matching degree of the current decision with the historical optimal solution in the feature space, and outputs a confidence score; the confidence score reflects the degree to which the current decision deviates from the training data distribution; the auxiliary verification module maps the main decision output to the historical robust decision region through feature space projection, determines whether it exceeds the credible interval, and if so, generates a correction coefficient based on the preset decision robustness rules; The confidence score output by the main decision module is weighted and fused with the correction coefficient to form the decision reliability index.

5. The real-time early warning algorithm for slope deformation based on deep reinforcement learning according to claim 2 is characterized in that: The process of marking the feature combinations to be optimized includes the following steps: The key feature weight vectors in the state quality assessment indicators are rescaled and standardized. Historical strategy failure cases are categorized according to their performance characteristics to establish a structured error event library. Each category of event is associated with a complete snapshot of the characteristic parameters at the time of the incident. A feature-error correlation matrix is constructed. By calculating the matching degree between feature vectors and error cases, the influence of each feature indicator on each type of error is quantified. The numerical value in the feature-error correlation matrix directly reflects the strength of the association between a specific feature and a certain type of error. The complete feature-error correlation matrix is spatially compressed and transformed to retain the dominant direction that explains most of the data variation, achieving the transformation from high-dimensional features to core features. In the feature space after spatial compression transformation, the main failure modes are screened according to the explanatory power of each dimension; the key dimensions obtained by the screening are reversely mapped back to the feature space to identify the most influential feature combination pattern that causes strategy failure.

6. The real-time early warning algorithm for slope deformation based on deep reinforcement learning according to claim 5 is characterized in that: Based on the failure mode analysis results, a feature improvement priority list is established. Through quantitative evaluation of the impact of each mode, the feature combinations that need to be optimized and their improvement order are clearly identified.

7. The real-time early warning algorithm for slope deformation based on deep reinforcement learning according to claim 1 is characterized in that: The process of identifying strategic weaknesses involves the following steps: Receive the strategy performance evaluation report. If the key indicators of the strategy performance evaluation report, such as false alarm rate, confidence matching degree, and feature drift index, exceed the preset security threshold, the vulnerability analysis mechanism will be triggered. Extract the standardized state representation vectors of all false positives, divide them into potential failure modes, compare the data distribution characteristics of each failure cluster, and screen out the key feature deviation patterns that are significantly correlated with the current strategy decision error; Calculate the cumulative deviation trend strength of historical monitoring data based on the characteristic drift index and predict the probability of future strategy failure; Based on potential failure modes and predicted future strategy failure probabilities, locate the feature set that causes a sudden drop in decision reliability. If the false alarm rate caused by a key feature changes significantly, the key feature is listed as the highest priority for optimization.

8. A real-time early warning system for slope deformation based on deep reinforcement learning, using the real-time early warning algorithm for slope deformation based on deep reinforcement learning as claimed in claim 1, characterized in that: Include: The indicator establishment module is used to integrate multi-source monitoring data of displacement, strain and groundwater level through a dynamic weighted fusion mechanism, screen the state characteristics of the multi-source monitoring data, generate standardized state representation vectors, and establish state quality assessment indicators; A report generation module receives the standardized state representation vector as input, builds a dual-mode policy architecture, generates preliminary decisions and their confidence assessments, and forms a quantitative indicator of decision reliability; Form final warning instructions and strategy update suggestions, and generate strategy performance evaluation reports; Reconstruction mechanism module, used to identify strategy weaknesses based on strategy performance evaluation reports; Continuously track the changing trends of state feature distribution, build a policy degradation warning system, and trigger the model reconstruction mechanism; the updated policy network parameters are fed back to the dual-mode policy architecture, and the optimized feature extraction suggestions are fed back to establish state quality assessment indicators.

Images